A common technique in communication systems, with unreliable and time-varying channel conditions is to correct errors based on automatic repeat request (ARQ) schemes together with a forward error correction (FEC) technique called hybrid ARQ (HARQ). If an error is detected by a commonly used cyclic redundancy check (CRC), the receiver of the communication system requests the transmitter to resend the erroneously received data packets.
S. Kallel, Analysis of a type II hybrid ARQ scheme with code combining, IEEE Transactions on Communications, Vol.38, No. 8, August 1990 and S. Kallel, R. Link, S. Bakhtiyari, Throughput performance of Memory ARQ schemes, IEEE Transactions on Vehicular Technology, Vol.48, No. 3, May 1999 define three different types of ARQ schemes:                Type I: The erroneous received packets are discarded and a new copy of the same packet is retransmitted and decoded separately. There is no combining of earlier and later received versions of that packet.        Type II: The erroneous received packets are not discarded, but are combined with some incremental redundancy bits provided by the transmitter for subsequent decoding. Retransmitted packets sometimes have higher coding rates and are combined at the receiver with the stored values. That means that only little redundancy is added in each retransmission.        Type III: Is the same as Type II with the constraint each retransmitted packet is now self-decodable. This implies that the transmitted packet is decodable without the combination with previous packets. This is useful if some packets are damaged in such a way that almost no information is reusable.        
Types II and III schemes are obviously more intelligent and show a performance gain with respect to Type I, because they provide the ability to reuse information from of previously received erroneous packets. There exist basically three schemes of reusing the redundancy of previously transmitted packets:                Soft-Combining        Code-Combining        Combination of Soft- and Code-CombiningSoft-Combining        
Employing soft-combining the retransmission packets carry identical symbols compared with the previously received symbols. In this case the multiple received packets are combined either by a symbol-by-symbol or by a bit-by-bit basis as for example disclosed in D. Chase, Code combining: A maximum-likelihood decoding approach for combining an arbitrary number of noisy packets, IEEE Trans. Commun., Vol. COM-33, pp. 385–393, May 1985 or B. A. Harvey and S. Wicker, Packet Combining Systems based on the Viterbi Decoder, IEEE Transactions on Communications, Vol. 42, No. 213/4, April 1994. By combining this soft-decision values from all received packets the reliabilities of the transmitted bits will increase linearly with the number and power of received packets. From a decoder point of view the same FEC scheme (with constant code rate) will be employed over all transmissions. Hence, the decoder does not need to know how many retransmissions have been performed, since it sees only the combined soft-decision values. In this scheme all transmitted packets will have to carry the same number of symbols.
Code-Combining
Code-combining concatenates the received packets in order to generate a new code word (decreasing code rate with increasing number of transmission). Hence, the decoder has to be aware of the FEC scheme to apply at each retransmission instant. Code-combining offers a higher flexibility with respect to soft-combining, since the length of the retransmitted packets can be altered to adapt to channel conditions. However, this requires more signaling data to be transmitted with respect to soft-combining.
Combination of Soft- and Code-Combining
In case the retransmitted packets carry some symbols identical to previously transmitted symbols and some code-symbols different from these, the identical code-symbols are combined using soft-combing as described in the section titled “Soft Combining” while the remaining code-symbols will be combined using code-combining. Here, the signaling requirements will be similar to code-combining.
As it has been shown in M. P. Schmitt, Hybrid ARQ Scheme employing TCM and Packet Combining, Electronics Letters Vol. 34, No. 18, September 1998 that HARQ performance for Trellis Coded Modulation (TCM) can be enhanced by rearranging the symbol constellation for the retransmissions. There, the performance gain results from the maximizing the Euclidean distances between the mapped symbols over the retransmissions, because the rearrangement has been performed on a symbol basis.
Considering high-order modulation schemes (with modulation symbols carrying more than two bits) the combining methods employing soft-combining have a major drawback: The bit reliabilities within soft-combined symbols will be in a constant ratio over all retransmissions, i.e. bits which have been less reliable from previous received transmissions will still be less reliable after having received further transmissions and, analogous, bits which have been more reliable from previous received transmissions will still be more reliable after having received further transmissions.
The varying bit reliabilities evolve from the constraint of two-dimensional signal constellation mapping, where modulation schemes-carrying more than 2 bits per symbol cannot have the same mean reliabilities for all bits under the assumption that all symbols are transmitted equally likely. The term mean reliabilities is consequently meant as the reliability of a particular bit over all symbols of a signal constellation.
Employing a signal constellation for a 16 QAM modulation scheme according to FIG. 1 showing a Gray encoded signal constellation with a given bit-mapping order i1q1i2q2, the bits mapped onto the symbols differ from each other in mean reliability in the first transmission of the packet. In more detail, bits i1 and q1 have a high mean reliability, as these bits are mapped to half spaces of the signal constellation diagram with the consequences that their reliability is independent from the fact of whether the bit transmits a one or a zero.
In contrast thereto, bits i2 and q2 have a low mean reliability, as their reliability depends on the fact of whether they transmit a one or a zero. For example, for bit i2, ones are mapped to outer columns, whereas zeros are mapped to inner columns. Similarly, for bit q2, ones are mapped to outer rows, whereas zeros are mapped to inner rows.
For the second and each further retransmissions the bit reliabilities will stay in a constant ratio to each other, which is defined by the signal constellation employed in the first transmission, i.e. bits i1 and q1 will always have a higher mean reliability than bits i2 and q2 after any number of retransmissions.